Detection of position of replaced mechanical heart valve with ANN

摘要

A mechanical heart valve is a device substituted for native heart valve. These valves are generally replaced with mitral or aorta valves. To found whether aorta or mitral heart valve replaced with mechanical one of patients is easy for specialist using stethoscope or chest X-ray. Expert system evaluating mechanical heart valve disease cannot deduce position of replaced heart valve. Thus, we aimed to determine which heart valve of patients was replaced with mechanical one to use preprocessing step in expert systems finding malfunctioning mechanical valve in this study. Electrocardiogram signal and 6 features extracted from power density of heart sounds were used to determine replaced heart valve using artificial neural networks. The heart sounds were separated into four parts according to recording area and sound component. Then, artificial neural networks was separately trained and tested for four sounds using the 6 features. As a result, the mechanical heart valve of patients was detected with 96.55% accuracy from the features of second heart sounds recorded from mitral area.